yet another dvcs shootout (this time python-devel)

[Robert Collins]

On Wed, 2008-11-05 at 10:53 +0000, Paul Moore wrote:
> 2008/11/5 Ollivier Robert <roberto at keltia.freenix.fr>:
> > According to Arne Bab.:
> >> $ time bzr branch http://code.python.org/python/3.0
> >> real    22m9.619s
> >> user    11m29.244s
> >> sys     0m14.302s
> >>
> >> $ time hg clone http://code.python.org/hg/branches/py3k/
> >> real    0m40.831s
> >> user    0m14.545s
> >> sys     0m3.896s
> >
> > I see that it has not been improved that much since the last time I played
> > with it...  This is still horribly slow...
> 
> My impression is that the Bazaar developers' answer would be that you
> should use a shared repository, and then branches after the first will
> be a lot faster. Personally, I find this argument bogus - I want the
> freedom to organise my repos as I prefer.

Shared repositories have nothing to do with initial clone speed. As a
Bazaar developer, I totally support you organising your repositories as
and how you desire. Slow initial clone speed is a serious problem  - one
we've done a lot of work on, with both a similar feature to hg's history
punching (but more similar to git's alternates), and direct performance
work. I just tried the bzr component of the above comparison with bzr's
1.8 release (about 5 months newer than 1.5):
Branched 37403
revision(s).                                                                                                                                                                                        

real    4m26.156s
user    3m56.170s
sys     0m11.020s

As already noted in this thread the two branches in question have very
different amounts of history - the hg one is well under 10% the depth of
the bzr one (and if only linear factors applied would excuse a ten-fold
increase in time). bzr 1.5 is missing a lot of performance work, and I
know that there were bugs, plain and simple, relating to index
processing and networking that caused non-linear scaling. (The bzr
branch goes back to 1990 - its cvs->svn->bzr).

Cheers,
Rob

-- 
GPG key available at: <http://www.robertcollins.net/keys.txt>.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : http://selenic.com/pipermail/mercurial/attachments/20081106/dc3e8172/attachment.pgp